Airborne Reflector-Based Ground Penetrating Radar for Environmental and Archaeological Studies

In this work, a wideband airborne reflector-based ground penetrating radar (GPR) is designed and proposed. The proposed GPR consists of 15 discrete panels, each panel supported by a single multi-rotor unmanned aerial vehicle (UAV). A lightweight, wideband Vivaldi antenna was developed as the feed structure. The antenna operates within the frequency range of 100–500 MHz and has a bandwidth of 387 MHz. The proposed antenna will possess a penetration depth of 2 m. Different panel configurations are studied and it is found that 15 discrete panels each with a dimension of 30 cm by 4 m and separated by 30 cm acts as a reflector antenna with a high gain of 14.25 dBi.


I. INTRODUCTION
I N THE fields of engineering and geophysics, Ground-Penetrating Radar (GPR) has a wide range of different applications including geology, archaeology, mineral exploration, glaciology, civil engineering, and more [1]. In [2] the application of ground-penetrating radar (GPR) for the exploration and characterization of limestone deposits has been discussed. The study demonstrates that the 3-D visualization of GPR data enhances the understanding of limestone distribution and facilitates more accurate reserve estimations. The findings highlight the potential of GPR as a useful tool for limestone exploration and mining operations. The application of ground-penetrating radar (GPR) in archaeology has been explored in [3]. It highlights two main aspects: the ability of GPR to map significant areas of archaeological sites that would otherwise remain invisible using traditional excavation methods, and the advancements in data acquisition and processing techniques. In [4] the use of Ground Penetrating Radar (GPR) in civil engineering and its potential applications in monitoring the earth and construction materials has been discussed. It emphasizes that GPR can effectively be utilized for civil engineering purposes, as the most commonly used construction materials such as concrete, asphalt, soils, and rocks are mineral aggregates that can be examined using GPR. A helicopter-borne GPR for environmental geology, and glaciology has been studied in [5]. The authors describe two types of radar systems: pulse radar and steppedfrequency radar (SFR). They explain that pulse radar systems have been used for measuring ice thickness, while SFR technology offers advantages such as reduced interference and high vertical resolution. The authors provide examples of survey data obtained using the pulse radar system, demonstrating its capability to measure bedrock reflections in thick Antarctic ice. They also present test results from the SFR system, conducted in different geological settings, including a military firing range and a glacier area in the Swiss Alps. However, to inspect and survey these rough, steep slopes, and hard-to-reach terrain, air-borne GPR plays a crucial role. By utilizing air-borne GPR, these challenging terrains will become more accessible. This technology will significantly enhance the survey process, making it more convenient and user-friendly.
Due to the wide range of GPR applications, various GPR antenna designs are commercially available, as well as utilized within the academic community for research purposes. GPR antenna type and size typically depend on the intended application. For instance, low frequency antennas that are larger in size are utilized when significant depth of penetration is crucial, while high frequency antennas that are smaller in size are employed when improved resolution and reduced penetration are necessary [6]. Airborne GPR is a prominent electromagnetic technology applied to identify surface or shallow subsurface features and for large-scale studies of hard-to-reach areas [7], [8], [9]. However, the most important challenge for this airborne GPR system is the antenna design. Airborne GPRs require antennas that have a wide bandwidth and high gain to effectively transmit and receive signals from elevated altitudes [7]. In [7] an ultrawideband (UWB) antipodal Vivaldi antenna is considered as an airborne GPR with a maximum gain of 11.5 dBi. Both the transmitter and the receiver antenna were installed on a wooden frame and hovered using a rope. The setup was placed 1.5 m above the ground and targeted object was buried 150 mm beneath the ground. A dual-polarized Vivaldi antenna array consisting of 8 × 8 elements for airborne radar measurements of snow was proposed in [8]. The antenna was mounted on a fixed-wing aircraft flown at an altitude of 500 m. In [9] different types of UWB Vivaldi antennas and their performance were analyzed in the context of an airborne-based GPR system for landmine and IED detection. The antennas were arranged in Rx-Tx-Rx configuration and were installed on a multi-rotor unmanned aerial vehicle (UAV). The prototype was placed 1.5 m above the ground with a maximum penetration depth of 18 cm.
The aim of this work is to develop an airborne multi-rotor UAV based lightweight broadband GPR antenna, operating at a lower frequency band with high gain. Moreover, we propose a novel idea to increase the gain of the antenna by implementing parabolic reflector composed of discrete panels which could be supported by the UAV payload, depicted in Fig. 1. The proposed antenna will be portable, usable on rough and inaccessible terrain, and capable of hovering for extended periods over an area of interest.
Reflector antennas have applications such as satellite communications, radio astronomy, remote sensing, radar, weaponry, and medical applications [10]. Among reflector antennas, the parabolic reflector is one of the most desirable because of its improved overall radiation characteristics [11] and it is also recognized as more compatible antenna for radiometry because it has lower sidelobes and higher main beam efficiencies [12].  In Section II we describe the design methodology of both the discrete parabolic reflector and the feed antenna. Results and the outputs are analyzed and discussed in Section III. Finally, in Section IV we conclude and discuss the future works.

A. DESIGN PROCEDURE OF THE DISCRETE PARABOLIC REFLECTOR
The design procedure for the center-fed parabolic reflector is comprehensively explained in [13]. The parabolic reflector design is shown in Fig. 2. The fundamental parameters that guide the design are the focal length, f , and the diameter of the parabola, D, depicted in Fig. 2. Initially, a parabolic reflector with an f /D ratio of 0.4 was designed based on equation (1) and the value of the D was kept 10 m and f of 4 m.
Afterward, in order to design a discrete parabolic reflector, 15 discrete rectangular metal panels were placed along the tangent of the continuous reflector. The distance between each of the rectangular sheets, P D was kept λ 0 /10, where λ 0 is the operating wavelength, here we considered 100 MHz as the minimum operating frequency. The panel reflectors are analogous to wire-grid reflectors, so the spacing between each of the rectangular panels was kept λ 0 /10 according to wire-grid reflector spacing rule [11]. Initially, the length, P L , and the width, P W , of each of the metal panels were kept 50 cm and 30 cm respectively as shown in Fig. 3. The first metal panel was placed at (0, 0) coordinates of the continuous parabola in x−z plane. To keep a distance of P D between the two adjacent metal panels, the x coordinate of the next panel along the continuous parabolic axis was calculated then  by substituting the value of x in equation (1) we determined the corresponding z coordinate. A detailed 2D positioning of 5 panels out of 15, along the continuous parabola is also illustrated in Fig. 3. For designed panel size of 30 cm by 50 cm, spacing, and the diameter of 10 m of the parabolic arc, a total of 15 panels is the upper limit on a single parabolic arc. Moreover, it is anticipated that a 15 UAV system represents a reasonable number of drones to control simultaneously. The limits of this control system would be further explored in future works. Detailed panel parameters are tabulated in Table 1.

B. FEEDING TECHNIQUES
The feeding structure for the parabolic reflector depends on various factors, such as the designed frequency, desired directivity, gain, polarization and the application requirements. In this work we designed a Vivaldi antenna as a feeding structure. The Vivaldi antenna was considered because we required a light weight antenna with broadband impedance matching, unidirectional radiation, compact size, and high directivity as GPR system requires a broadband antenna with high gain [7]. Moreover, this airborne reflector based GPR requires a lightweight antenna to ensure maximum flight time and minimal payload.
The geometry of the proposed feed antenna is depicted in Fig. 4. The antenna was designed to have a minimum operating frequency of 100 MHz and a maximum operating frequency was considered 500 MHz. This frequency band was chosen according to FCC guidelines for GPR operation below 960 MHz [14], [15]. The substrate is FR4 with a dielectric constant, r of 4.3, and loss tangent, tan δ, of 0.025. The FR4 substrate has a thickness of 3.2 mm. The exponentially tapered slots can be determined by using  [16].
In equation (2) R is the opening rate of the profile which was considered at 10.5, and K 1 and K 2 are determined by the coordinates of the starting and end point of the exponential curve using the equations (3) and (4), described below [16].
According to [17], in order to radiate electromagnetic energy from the taper slots, the separation between the two flares needs to be greater than λ c /2, where λ c is the center frequency of the whole operating bandwidth. So, the starting and end point was chosen in such a way so that the taper length T L and the taper width, T W , is greater than λ c /2.
The lower frequency limit is primarily determined by the antenna's width and the effective dielectric constant of the substrate, and can be determined using the equation (5) below [17] Here, f 0 , is the minimum operating frequency and eff is the effective dielectric constant of the substrate. The taper slot was terminated by a circular slot with a diameter, d, equivalent to one quarter-wavelength at the center frequency to further broaden the bandwidth. A microstrip to slotline transition feeding structure was designed to feed the antenna using a three-stage quarter-wave balun feeding. The feeding structure was designed using binomial multisection matching quarter-wave transformer. The initial feeding microstrip line was designed to have an input impedance of 50 and the corresponding width, w 1 , was kept at 6 mm. Then the threestage multisection transformer was designed for improved impedance transition. The width of each of the multisection w 2 , w 3 , w 4 were set to 5.4 mm, 3.5 mm, and 0.66 mm corresponding to 55.5 , 70 , and 130.5 , respectively. The length of each of the microstrip was kept λ m /4, where λ m is the microstrip guided wavelength. Moreover, to have a better impedance matching bandwidth the feeding microstrip line was also extended by a quarter-wavelength short circuit on the left side of the feeding line [18]. The optimized designed parameters are in Table 3.

III. RESULT ANALYSIS
The Vivaldi antenna was designed and simulated using CST Microwave Studio Suite. Fig. 5 shows the reflection   Fig. 6. From Fig. 6 it can be observed that by implementing the reflector the bandwidth of the antenna decreases. For 50 cm, 1 m, 2 m, and 4 m of panel length, the antenna has a −10 dB bandwidth of 40.5 MHz, 42.3 MHz, 40 MHz, and 110 MHz respectively. As seen from the S 11 plot, the panels of length less than 4 m exhibit very similar S 11 characteristics. The reason for this is further explored by investigating the far-field radiation pattern in the coming paragraphs.
The parabolic reflector gain improvement is clearly seen in Fig. 7 which shows the gain comparison of the Vivaldi antenna with and without the reflector panels. Fig. 7 shows that by deploying the reflector metal panel the gain of  The far-field radiation pattern for each of the reflector panels at 425 MHz is shown in Fig. 8. Fig. 8 shows that the panels with lengths of 50 cm, 1 m, and 2 m are not working as a reflector, from the location of maximum (red) pattern we can see the peak radiation pattern has not been reflected by the panels. This also explains why the gain in Fig. 6 is very similar for panels of lengths 0.5, 1, and 2 m. For these panels, the peak radiation pattern is in the direction of maximum gain of the feed Vivaldi antenna. For these dimensions, the panels act more like director elements which incidentally provide some gain enchantment. However, the panels with a length of 4 m work as both a reflector and as well as gain enhancer. Thus, although the gain is increasing for each panel length in Fig. 7, the direction of the gain is not the same.
The phenomenon observed in Figs. 7 and 8 is because the gain and directivity of a reflector antenna is directly proportional to its aperture efficiency and this aperture efficiency is a combination of aperture taper efficiency and spillover efficiency. When the dimension of the panels is smaller, the aperture taper efficiency and spillover efficiency is low, as a consequence the panels with smaller dimension is not acting like a reflector antenna. However, with a larger panel dimension, the efficiency is high and as a result, the proposed antenna works as a reflector antenna. Figs. 9 and 10 depict the radiation pattern in the E-plane and H-plane, respectively, for each of the reflector panels operating at a frequency of 425 MHz. For a panel length of 50 cm, the E-plane radiation pattern exhibits a main lobe direction at 144 • with a 3 dB beamwidth of 29.6 • . For 1 m and 2 m, the main lobe Moreover, we can also assert that as our antenna is operating well within the frequency band from 200 MHz to 500 MHz, the proposed antenna is able to obtain a maximum penetration depth of 2 m [19]. This proposed antenna will also have a high resolution as the resolution of the GPR antenna is directly proportional to its bandwidth [20].
A comprehensive comparative analysis has been done to compare the performance of the proposed GPR antenna with the pre-existing GPR antennas, presented in Table 3. The performance of the proposed antenna exhibits promising results when compared to the reported performances of antennas in the existing literature. Although the proposed antenna has a lower bandwidth compared to the other related works. However, the gain of the proposed antenna is comparatively high than those of pre-existing GPR antennas. Moreover, compared with other air-borne GPR antennas, the proposed antenna also exhibits a higher gain.

IV. CONCLUSION AND FUTURE WORKS
In this work, a reflector based airborne GPR antenna is proposed. The antenna has a high gain and possesses a high penetration depth of 2 m. A discrete parabolic reflector was proposed and a wideband, lightweight Vivaldi antenna was designed as a feeding structure. The operational frequency of the proposed antenna is 100-500 MHz and the antenna has a bandwidth of 387 MHz. The discrete parabolic reflector with a dimension of 30 cm by 4 m works as a reflector antenna with a maximum gain of 14.25 dBi. Our proposed antenna will also have a better resolution in the lower frequency band due to its wideband characteristics. In future works, the antenna panel configuration will be further optimized for integration into a multi-rotor UAV system. Moreover, the maximum flight time of the UAV with payload and its' control system to maintain the desired spacing between the reflectors will also be investigated in the future. Additionally, the whole prototype will be fabricated in the future to further validated our proposed work.